Skip to content

From Reads to a Matrix

Terminal window
nextflow run nf-core/scrnaseq \
-profile singularity \
--input samplesheet.csv \
--outdir results \
--genome GRCh38 \
--aligner cellranger
Terminal window
cellranger count \
--id=sample1 \
--transcriptome=/ref/refdata-gex-GRCh38 \
--fastqs=/data/fastqs \
--sample=sample1
sample1/
├── outs/
│ ├── filtered_feature_bc_matrix/ # Use this
│ │ ├── barcodes.tsv.gz
│ │ ├── features.tsv.gz
│ │ └── matrix.mtx.gz
│ ├── raw_feature_bc_matrix/ # Includes empty droplets
│ ├── web_summary.html # QC report
│ └── cloupe.cloupe # Loupe Browser file
Cell_1 Cell_2 Cell_3 ...
Gene_A 0 5 0
Gene_B 12 0 3
Gene_C 0 0 0
...
  • Rows: Genes (~20,000)
  • Columns: Cells (1,000-100,000)
  • Values: UMI counts (mostly zeros!)

Standard tutorial dataset from 10x Genomics:

  • Peripheral blood mononuclear cells
  • Well-characterized cell types
  • Available at: 10x datasets
# Load PBMC 3k directly
library(SeuratData)
InstallData("pbmc3k")
data("pbmc3k")

Mouse organ atlas:

  • Multiple tissues
  • Both 10x and Smart-seq2
  • Great for reference mapping

Comprehensive human reference: